Method and system for evaluating stability of power system

Abstract

The invention discloses a stability evaluation method and system for a power system, and the method comprises the steps: combining a first node feature of each device in the power system with a first edge feature containing an incidence relation to form a graph model, the first node feature comprising a frequency, a voltage, a phase and a power angle; based on a to-be-determined stability type of the power system, extracting a corresponding sub-graph from the graph model to form a corresponding stable sub-graph; extracting node features in the stability sub-graph through a preset graph neural network model, and judging the node features to obtain a stability probability corresponding to the power system; and if the stability probability does not meet the preset condition, performing stability evaluation on the power system through a preset graph calculation stability judgment algorithm. The method can improve the accuracy and efficiency of power system stability evaluation, and is suitable for large-scale power grid stability analysis.

Description

Technical Field

[0001] The present application relates to the field of power system stability analysis, and in particular to a power system stability assessment method and system. Background Art

[0002] With the increasing proportion of new energy in new power systems and the reduction of system inertia, the operation mode of power systems has become more volatile, which has brought many challenges to their safe and stable operation.

[0003] Traditional methods for stability assessment of power systems include direct methods and indirect methods. The direct method has the characteristics of fast calculation speed and online application, but it usually simplifies the system model, resulting in poor assessment accuracy. The indirect method relies on transient simulation calculations and has the characteristics of slow calculation speed and high accuracy, but because it often requires a complex calculation process, it has disadvantages such as low calculation efficiency. For large-scale power systems or scenarios with high real-time requirements, the calculation time of the indirect method will become a bottleneck for its application.

[0004] Application Contents

[0005] The present application provides a method and system for evaluating the stability of an electric power system to improve the accuracy and efficiency of evaluating the stability of an electric power system, which is suitable for large-scale power grid stability analysis.

[0006] In a first aspect, the present application provides a method for evaluating the stability of a power system, comprising:

[0007] A graph model is formed based on the first node feature of each device in the power system and the first edge feature containing the association relationship, wherein the first node feature includes: frequency, voltage, phase and power angle;

[0008] Based on the stability type to be determined of the power system, extracting a corresponding subgraph from the graph model to form a corresponding stable subgraph;

[0009] The node features in the stability subgraph are extracted by a preset graph neural network model, and the node features are judged to obtain the stability probability corresponding to the power system;

[0010] If the stability probability does not meet the preset conditions, the stability of the power system is evaluated by using a preset graph calculation stability judgment algorithm.

[0011] The present application can accurately and quickly extract corresponding subgraphs from a preset graph model according to different stability judgment types, so as to facilitate the subsequent judgment of the stability probability of the power system and the calculation of the stability of the power system; the preset graph neural network model can effectively capture the node features in the stability subgraph, and can give full play to the advantages of the fast calculation of the graph neural network model, so as to quickly calculate the stability probability of the power system; through the graph calculation stability judgment algorithm, the node features whose stability probabilities do not meet the preset conditions are deeply analyzed, so as to more accurately and comprehensively evaluate the stability of the entire power system, so as to improve the accuracy and efficiency of the stability evaluation of the power system, which is suitable for large-scale power grid stability analysis.

[0012] Furthermore, the construction process of the graph model is specifically as follows:

[0013] Acquire the first node feature of each device in the power system and combine it with the first edge feature containing the association relationship to form a basic graph model;

[0014] Adding mutually connected interface nodes, differential nodes and algebraic nodes to the basic graph model to form a graph model; wherein the interface nodes are connected to the basic graph model.

[0015] In this way, a basic graph model can be constructed according to the correlation between the node data and edge data of the power system, and interface nodes, differential nodes and algebraic nodes can be added to the basic graph model. This can facilitate the extraction of corresponding subgraphs from the preset graph model based on different needs, and facilitate the subsequent stability assessment of the power system through the graph calculation stability judgment algorithm.

[0016] Furthermore, based on the stability type to be determined of the power system, a corresponding subgraph is extracted from the graph model to form a corresponding stable subgraph, specifically:

[0017] Based on the stability type to be determined of the power system, reading the second node feature and the second edge feature corresponding to each device in the graph model;

[0018] And based on the second node feature and the second edge feature, a corresponding subgraph is extracted from the graph model to form the corresponding stable subgraph.

[0019] In this way, the corresponding subgraphs can be accurately and quickly extracted from the preset graph model according to different stability judgment types, so as to facilitate the subsequent judgment of stability probability and stability evaluation.

[0020] Furthermore, the node features in the stability subgraph are extracted by using a preset graph neural network model, and the node features are judged to obtain the stability probability corresponding to the power system, specifically:

[0021] Inputting the stable subgraph into the graph neural network model;

[0022] Extracting a third node feature in the stable subgraph by using the graph neural network model, and normalizing the third node feature to obtain a node stability probability corresponding to the third node feature;

[0023] By calculating the node stability probabilities corresponding to all nodes in the power system in a preset time period, the corresponding stability probability of the power system is obtained.

[0024] In this way, the preset graph neural network model can effectively capture the node features in the stable subgraph, and can give full play to the advantages of fast calculation of the graph neural network model, so as to quickly calculate the stability probability of the power system.

[0025] Furthermore, the stability probability corresponding to the power system is obtained as follows:

[0026]

[0027] In the formula, f( x ) is the corresponding stability probability of the power system, N is the number of nodes in the power system, N t For custom time periods, For the i The node stability probability of a node at time t.

[0028] In this way, by calculating the stability probability corresponding to each node, the stability probability of the entire power system can be obtained.

[0029] In a second aspect, the present application provides a stability assessment system for a power system, comprising: a construction module, an extraction module, a first processing module, and a second processing module;

[0030] The construction module is used to form a graph model based on the first node characteristics of each device in the power system and the first edge characteristics containing the association relationship, wherein the first node characteristics include: frequency, voltage, phase and power angle;

[0031] The extraction module is used to extract the corresponding subgraph from the graph model based on the stability type to be determined of the power system to form a corresponding stable subgraph;

[0032] The first processing module is used to extract node features in the stability subgraph through a preset graph neural network model, and judge the node features to obtain the stability probability corresponding to the power system;

[0033] The second processing module is used to evaluate the stability of the power system by using a preset graph calculation stability judgment algorithm if the stability probability does not meet a preset condition.

[0034] The present application can accurately and quickly extract corresponding subgraphs from a preset graph model according to different stability judgment types, so as to facilitate the subsequent judgment of the stability probability of the power system and the calculation of the stability of the power system; the preset graph neural network model can effectively capture the node features in the stability subgraph, and can give full play to the advantages of the fast calculation of the graph neural network model, so as to quickly calculate the stability probability of the power system; through the graph calculation stability judgment algorithm, the node features whose stability probabilities do not meet the preset conditions are deeply analyzed, so as to more accurately and comprehensively evaluate the stability of the entire power system, so as to improve the accuracy and efficiency of the stability evaluation of the power system, which is suitable for large-scale power grid stability analysis.

[0035] Furthermore, the construction process of the graph model is specifically as follows:

[0036] Acquire the first node feature of each device in the power system and combine it with the first edge feature containing the association relationship to form a basic graph model;

[0037] Adding mutually connected interface nodes, differential nodes and algebraic nodes to the basic graph model to form a graph model; wherein the interface nodes are connected to the basic graph model.

[0038] Further, the extraction module includes: a reading unit and a forming unit;

[0039] The reading unit is used to read the second node feature and the second edge feature corresponding to each device in the graph model based on the stability type to be determined of the power system;

[0040] The forming unit is used to extract a corresponding subgraph from the graph model based on the second node feature and the second edge feature to form the corresponding stable subgraph.

[0041] Further, the first processing module includes: an input unit, a first processing unit and a second processing unit;

[0042] The input unit is used to input the stable subgraph into the graph neural network model;

[0043] The first processing unit is used to extract the third node feature in the stable subgraph through the graph neural network model, and normalize the third node feature to obtain the node stability probability corresponding to the third node feature;

[0044] The second processing unit is used to obtain the stability probability corresponding to the power system by calculating the node stability probability corresponding to all nodes in the power system in a preset time period.

[0045] Further, the obtaining of the stability probability of the power system is specifically:

[0046]

[0047] In the formula, f( x ) is the corresponding stability probability of the power system, N is the number of nodes in the power system, N t For custom time periods, For the i The node stability probability of a node at time t. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a flow chart of an embodiment of a method for evaluating the stability of a power system provided by the present application;

[0049] Figure 2 It is a structural schematic diagram of an embodiment of a graph model provided by the present application;

[0050] Figure 3 is a structural schematic diagram of another embodiment of a graph model provided by the present application;

[0051] Figure 4 It is a structural schematic diagram of an embodiment of a stability assessment system for a power system provided in the present application. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0053] It should be understood that the step numbers used in this article are only for the convenience of description and are not intended to limit the order in which the steps are executed.

[0054] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an" and "the" are intended to include plural forms unless the context clearly indicates otherwise.

[0055] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0056] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0057] As the proportion of renewable energy in the power system increases and the system inertia decreases, the operation of the power system becomes more volatile, bringing many challenges to safe and stable operation. Traditional methods for evaluating the stability of power systems include direct and indirect methods. The direct method is fast to calculate, but has poor accuracy due to the simplified model. The indirect method relies on transient simulation, which has high accuracy but slow calculation speed, and is difficult to apply to large-scale power systems or scenarios with high real-time requirements.

[0058] Embodiment 1

[0059] Please refer to Figure 1 , Figure 1 It is a flow chart of an embodiment of a method for evaluating the stability of a power system provided by the present application, comprising steps S1 to S4;

[0060] Step S1, forming a graph model according to the first node feature of each device in the power system and the first edge feature containing the association relationship, wherein the first node feature includes: frequency, voltage, phase and power angle;

[0061] It can be understood that the first node feature is a functional feature of various devices in the power system, such as frequency, voltage, phase and power angle, etc., and the first edge feature is the connection relationship between the first node features. With the popularization of power system measurement equipment, the measurement information of the above-mentioned various equipment components in the power system can be obtained in real time, that is, the node features and edge features of each equipment component. The basic graph model can be formed by combining the node features and the edge features.

[0062] In addition, the basic graph model may be added with interconnected interface nodes, differential nodes and algebraic nodes to form a graph model; wherein the interface nodes are connected to the basic graph model. Specifically, a structural diagram of an embodiment of the graph model is as follows: Figure 2As shown, the interface node 21, the differential node 22 and the algebraic node 23 are connected in pairs to form a triangle, each dotted triangle part forms the transient simulation calculation part of a single device element, and the transient simulation calculation parts of different device elements are independent of each other. The node part represented by the dotted edge and the double circle is used to describe the dynamic equation of the element, and is uniquely connected to the basic graph model through the interface node. The attributes of the interface node 21 include the voltage and current at the interface, the attributes of the differential node 22 include the differential state quantity and related parameters of the device element, and the attributes of the algebraic node 23 include the algebraic variables and related parameters of the device element. By adding interconnected interface nodes, differential nodes and algebraic nodes to each device in the power system, a graph model can be formed.

[0063] It should be noted that the first node feature can be a spatial feature such as voltage, frequency and power angle, or a time feature. There is an abstract edge between the time feature and each device, and the edge attribute is the measurement information of the device at the time node.

[0064] In this way, a basic graph model can be constructed according to the correlation between the node data and edge data of the power system, and interface nodes, differential nodes and algebraic nodes can be added to the basic graph model. This can facilitate the extraction of corresponding subgraphs from the preset graph model based on different needs, and facilitate the subsequent stability assessment of the power system through the graph calculation stability judgment algorithm.

[0065] Step S2: based on the stability type to be determined of the power system, extract a corresponding subgraph from the graph model to form a corresponding stable subgraph;

[0066] It can be understood that based on the stability type to be determined of the power system, the second node feature and the second edge feature corresponding to each device in the graph model are read through the neural network; and based on the second node feature and the second edge feature, the corresponding subgraph is extracted from the graph model to form the corresponding stable subgraph, and the stable subgraph is stored in the memory to support the determination of the subsequent graph neural network model. The structural schematic diagram of another embodiment of the graph model is shown in FIG. Figure 3As shown. Exemplarily, if the stability type to be determined is frequency stability determination, the frequency information corresponding to each device in the graph model and the associated information related to the frequency are read, and the corresponding subgraph is extracted from the graph model based on the frequency information and the frequency associated information to form a frequency stability determination subgraph; if the stability type to be determined is voltage stability, the voltage information corresponding to each device in the graph model and the associated information related to the voltage are read, and the corresponding subgraph is extracted from the graph model based on the voltage information and the voltage associated information to form a voltage stability determination subgraph; if the stability type to be determined is power angle stability, the power angle of the generator in the graph model and the associated information related to the power angle are read, and the corresponding subgraph is extracted from the graph model based on the power angle information and the power angle associated information to form a power angle stability determination subgraph.

[0067] It should be noted that, according to different stability types, multiple neural networks can be trained separately, or one neural network can be trained to extract all subgraphs. This application does not limit the number of neural networks.

[0068] In this way, the corresponding subgraphs can be accurately and quickly extracted from the preset graph model according to different stability judgment types, so as to facilitate the subsequent judgment of stability probability and stability evaluation.

[0069] Step S3, extracting node features in the stability subgraph through a preset graph neural network model, and judging the node features to obtain the stability probability corresponding to the power system;

[0070] It can be understood that the stable subgraph is input into the graph neural network model; the third node feature in the stable subgraph is extracted by the graph neural network model, and the third node feature is normalized to obtain the node stability probability corresponding to the third node feature; the node stability probability corresponding to all nodes in the power system in the preset time period is calculated to obtain the stability probability corresponding to the power system. Specifically, the stable subgraph is input into the graph neural network model, and the relationship and local features between the nodes in the stable subgraph are first extracted through each layer of the point cloud dynamic graph convolution (DGCNN) network to construct a K-nearest neighbor graph for each node, and the similar nodes are gathered together; then, the edge convolution is performed on the K-nearest neighbor graph through the graph convolution network (GCN) to further extract the global features of the node; again, the local features and the global features are fused through the fully connected network (MLP) to obtain the node features, and the node features are normalized to obtain the node stability probability corresponding to the node features, and the node stability probability corresponding to the node features is calculated by calculating the node stability probability of all nodes in the power system, and the stability probability corresponding to the power system can be obtained.

[0071] The stability probability corresponding to the power system is obtained as follows:

[0072]

[0073] In the formula, f( x ) is the corresponding stability probability of the power system, N is the number of nodes in the power system, N t For custom time periods, For the i The node stability probability of a node at time t.

[0074] It should be noted that in order to enable the graph neural network model to adaptively focus on key node features and edge features, an attention layer is introduced based on the graph convolutional network (GCN).

[0075] In this way, the preset graph neural network model can effectively capture the node features in the stable subgraph, and can give full play to the advantages of fast calculation of the graph neural network model, so as to quickly calculate the stability probability of the power system.

[0076] Step S4: If the stability probability does not meet the preset conditions, the stability of the power system is evaluated by using a preset graph calculation stability judgment algorithm.

[0077] It can be understood that when the stability probability meets the preset conditions, the power system stability assessment results can be output; however, when the stability probability does not meet the preset conditions, it is necessary to further conduct stability assessment of the power system through the preset graph calculation stability judgment algorithm, and then the power system stability assessment results can be output.

[0078] It should be noted that the preset conditions can be freely set according to needs and this application does not impose any restrictions.

[0079] The present application can accurately and quickly extract corresponding subgraphs from a preset graph model according to different stability judgment types, so as to facilitate the subsequent judgment of the stability probability of the power system and the calculation of the stability of the power system; the preset graph neural network model can effectively capture the node features in the stability subgraph, and can give full play to the advantages of the fast calculation of the graph neural network model, so as to quickly calculate the stability probability of the power system; through the graph calculation stability judgment algorithm, the node features whose stability probabilities do not meet the preset conditions are deeply analyzed, so as to more accurately and comprehensively evaluate the stability of the entire power system, so as to improve the accuracy and efficiency of the stability evaluation of the power system, which is suitable for large-scale power grid stability analysis.

[0080] Embodiment 2

[0081] Please refer to Figure 4 , Figure 4 It is a structural schematic diagram of an embodiment of a stability assessment system for a power system provided by the present application, comprising a construction module 01, an extraction module 02, a first processing module 03 and a second processing module 04;

[0082] The construction module 01 is used to form a graph model based on the first node characteristics of each device in the power system and the first edge characteristics containing the association relationship, wherein the first node characteristics include: frequency, voltage, phase and power angle;

[0083] It can be understood that the first node feature is a functional feature of various devices in the power system, such as frequency, voltage, phase and power angle, etc., and the first edge feature is the connection relationship between the first node features. With the popularization of power system measurement equipment, the measurement information of the above-mentioned various equipment components in the power system can be obtained in real time, that is, the node features and edge features of each equipment component. The basic graph model can be formed by combining the node features and the edge features.

[0084] In addition, the basic graph model may be added with interconnected interface nodes, differential nodes and algebraic nodes to form a graph model; wherein the interface nodes are connected to the basic graph model. Specifically, a structural diagram of an embodiment of the graph model is as follows: Figure 2 As shown, the interface node 21, the differential node 22 and the algebraic node 23 are connected in pairs to form a triangle, each dotted triangle part forms the transient simulation calculation part of a single device element, and the transient simulation calculation parts of different device elements are independent of each other. The node part represented by the dotted edge and the double circle is used to describe the dynamic equation of the element, and is uniquely connected to the basic graph model through the interface node. The attributes of the interface node 21 include the voltage and current at the interface, the attributes of the differential node 22 include the differential state quantity and related parameters of the device element, and the attributes of the algebraic node 23 include the algebraic variables and related parameters of the device element. By adding interconnected interface nodes, differential nodes and algebraic nodes to each device in the power system, a graph model can be formed.

[0085] It should be noted that the first node feature can be a spatial feature such as voltage, frequency and power angle, or a time feature. There is an abstract edge between the time feature and each device, and the edge attribute is the measurement information of the device at the time node.

[0086] In this way, a basic graph model can be constructed according to the correlation between the node data and edge data of the power system, and interface nodes, differential nodes and algebraic nodes can be added to the basic graph model. This can facilitate the extraction of corresponding subgraphs from the preset graph model based on different needs, and facilitate the subsequent stability assessment of the power system through the graph calculation stability judgment algorithm.

[0087] The extraction module 02 is used to extract the corresponding subgraph from the graph model based on the stability type to be determined of the power system to form a corresponding stable subgraph;

[0088] Specifically, the extraction module 02 includes: a reading unit and a forming unit; the reading unit is used to read the second node characteristics and the second edge characteristics corresponding to each device in the graph model based on the stability type to be determined of the power system; the forming unit is used to extract the corresponding subgraph from the graph model based on the second node characteristics and the second edge characteristics to form the corresponding stable subgraph.

[0089] It can be understood that based on the stability type to be determined of the power system, the second node feature and the second edge feature corresponding to each device in the graph model are read through the neural network; and based on the second node feature and the second edge feature, the corresponding subgraph is extracted from the graph model to form the corresponding stable subgraph, and the stable subgraph is stored in the memory to support the determination of the subsequent graph neural network model. The structural schematic diagram of another embodiment of the graph model is shown in FIG. Figure 3 As shown. Exemplarily, if the stability type to be determined is frequency stability determination, the frequency information corresponding to each device in the graph model and the associated information related to the frequency are read, and the corresponding subgraph is extracted from the graph model based on the frequency information and the frequency associated information to form a frequency stability determination subgraph; if the stability type to be determined is voltage stability, the voltage information corresponding to each device in the graph model and the associated information related to the voltage are read, and the corresponding subgraph is extracted from the graph model based on the voltage information and the voltage associated information to form a voltage stability determination subgraph; if the stability type to be determined is power angle stability, the power angle of the generator in the graph model and the associated information related to the power angle are read, and the corresponding subgraph is extracted from the graph model based on the power angle information and the power angle associated information to form a power angle stability determination subgraph.

[0090] It should be noted that, according to different stability types, multiple neural networks can be trained separately, or one neural network can be trained to extract all subgraphs. This application does not limit the number of neural networks.

[0091] In this way, the corresponding subgraphs can be accurately and quickly extracted from the preset graph model according to different stability judgment types, so as to facilitate the subsequent judgment of stability probability and stability evaluation.

[0092] The first processing module 03 is used to extract node features in the stability subgraph through a preset graph neural network model, and judge the node features to obtain the stability probability corresponding to the power system;

[0093] Specifically, the first processing module 03 includes: an input unit, a first processing unit and a second processing unit; the input unit is used to input the stable subgraph into the graph neural network model; the first processing unit is used to extract the third node feature in the stable subgraph through the graph neural network model, and normalize the third node feature to obtain the node stability probability corresponding to the third node feature; the second processing unit is used to obtain the stability probability corresponding to the power system by calculating the node stability probability corresponding to all nodes in the power system in a preset time period.

[0094] It can be understood that the stable subgraph is input into the graph neural network model; the third node feature in the stable subgraph is extracted by the graph neural network model, and the third node feature is normalized to obtain the node stability probability corresponding to the third node feature; the node stability probability corresponding to all nodes in the power system in the preset time period is calculated to obtain the stability probability corresponding to the power system. Specifically, the stable subgraph is input into the graph neural network model, and the relationship and local features between the nodes in the stable subgraph are first extracted through each layer of the point cloud dynamic graph convolution (DGCNN) network to construct a K-nearest neighbor graph for each node, and the similar nodes are gathered together; then, the edge convolution is performed on the K-nearest neighbor graph through the graph convolution network (GCN) to further extract the global features of the node; again, the local features and the global features are fused through the fully connected network (MLP) to obtain the node features, and the node features are normalized to obtain the node stability probability corresponding to the node features, and the node stability probability corresponding to the node features is calculated by calculating the node stability probability of all nodes in the power system, and the stability probability corresponding to the power system can be obtained.

[0095] The stability probability corresponding to the power system is obtained as follows:

[0096]

[0097] In the formula, f( x ) is the corresponding stability probability of the power system, N is the number of nodes in the power system, N t For custom time periods, For the iThe node stability probability of a node at time t.

[0098] It should be noted that in order to enable the graph neural network model to adaptively focus on key node features and edge features, an attention layer is introduced based on the graph convolutional network (GCN).

[0099] In this way, the preset graph neural network model can effectively capture the node features in the stable subgraph, and can give full play to the advantages of fast calculation of the graph neural network model, so as to quickly calculate the stability probability of the power system.

[0100] The second processing module 04 is used to evaluate the stability of the power system by using a preset graph calculation stability judgment algorithm if the stability probability does not meet a preset condition.

[0101] It can be understood that when the stability probability meets the preset conditions, the power system stability assessment results can be output; however, when the stability probability does not meet the preset conditions, it is necessary to further conduct stability assessment of the power system through the preset graph calculation stability judgment algorithm, and then the power system stability assessment results can be output.

[0102] It should be noted that the preset conditions can be freely set according to needs and this application does not impose any restrictions.

[0103] The present application can accurately and quickly extract corresponding subgraphs from a preset graph model according to different stability judgment types, so as to facilitate the subsequent judgment of the stability probability of the power system and the calculation of the stability of the power system; the preset graph neural network model can effectively capture the node features in the stability subgraph, and can give full play to the advantages of the fast calculation of the graph neural network model, so as to quickly calculate the stability probability of the power system; through the graph calculation stability judgment algorithm, the node features whose stability probabilities do not meet the preset conditions are deeply analyzed, so as to more accurately and comprehensively evaluate the stability of the entire power system, so as to improve the accuracy and efficiency of the stability evaluation of the power system, which is suitable for large-scale power grid stability analysis.

[0104] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the scope of protection of the present application. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of protection of the present application.

Claims

1. A method for evaluating the stability of a power system, characterized in that: include: A graph model is formed based on the first node feature of each device in the power system and the first edge feature containing the association relationship, wherein the first node feature includes: frequency, voltage, phase and power angle; Based on the stability type to be determined of the power system, extracting a corresponding subgraph from the graph model to form a corresponding stable subgraph; The node features in the stability subgraph are extracted by a preset graph neural network model, and the node features are judged to obtain the stability probability corresponding to the power system; If the stability probability does not meet the preset conditions, the stability of the power system is evaluated by using a preset graph calculation stability judgment algorithm.

2. The method for evaluating the stability of a power system according to claim 1, characterized in that: The construction process of the graph model is specifically as follows: Acquire the first node feature of each device in the power system and combine it with the first edge feature containing the association relationship to form a basic graph model; Adding mutually connected interface nodes, differential nodes and algebraic nodes to the basic graph model to form a graph model; wherein the interface nodes are connected to the basic graph model.

3. The method for evaluating the stability of a power system according to claim 1, characterized in that: Based on the stability type to be determined of the power system, a corresponding subgraph is extracted from the graph model to form a corresponding stable subgraph, specifically: Based on the stability type to be determined of the power system, reading the second node feature and the second edge feature corresponding to each device in the graph model; And based on the second node feature and the second edge feature, a corresponding subgraph is extracted from the graph model to form the corresponding stable subgraph.

4. The method for evaluating the stability of a power system according to claim 1, characterized in that: The node features in the stable subgraph are extracted by a preset graph neural network model, and the node features are judged to obtain the stability probability corresponding to the power system, specifically: Inputting the stable subgraph into the graph neural network model; Extracting a third node feature in the stable subgraph by using the graph neural network model, and normalizing the third node feature to obtain a node stability probability corresponding to the third node feature; By calculating the node stability probabilities corresponding to all nodes in the power system in a preset time period, the corresponding stability probability of the power system is obtained.

5. The method for evaluating the stability of a power system according to any one of claims 1 or 4, characterized in that: The stability probability corresponding to the power system is obtained as follows: In the formula, f( x ) is the corresponding stability probability of the power system, N is the number of nodes in the power system, N t For custom time periods, For the i The node stability probability of a node at time t.

6. A stability assessment system for a power system, characterized in that: include: A construction module, an extraction module, a first processing module, and a second processing module; The construction module is used to form a graph model based on the first node characteristics of each device in the power system and the first edge characteristics containing the association relationship, wherein the first node characteristics include: frequency, voltage, phase and power angle; The extraction module is used to extract the corresponding subgraph from the graph model based on the stability type to be determined of the power system to form a corresponding stable subgraph; The first processing module is used to extract node features in the stability subgraph through a preset graph neural network model, and judge the node features to obtain the stability probability corresponding to the power system; The second processing module is used to evaluate the stability of the power system by using a preset graph calculation stability judgment algorithm if the stability probability does not meet a preset condition.

7. The stability assessment system of the power system according to claim 6, characterized in that: The construction process of the graph model is specifically as follows: Acquire the first node feature of each device in the power system and combine it with the first edge feature containing the association relationship to form a basic graph model; Adding mutually connected interface nodes, differential nodes and algebraic nodes to the basic graph model to form a graph model; wherein the interface nodes are connected to the basic graph model.

8. The stability assessment system of the power system according to claim 6, characterized in that: The extraction module comprises: a reading unit and a forming unit; The reading unit is used to read the second node feature and the second edge feature corresponding to each device in the graph model based on the stability type to be determined of the power system; The forming unit is used to extract a corresponding subgraph from the graph model based on the second node feature and the second edge feature to form the corresponding stable subgraph.

9. The stability assessment system of the power system according to claim 6, characterized in that: The first processing module includes: an input unit, a first processing unit and a second processing unit; The input unit is used to input the stable subgraph into the graph neural network model; The first processing unit is used to extract the third node feature in the stable subgraph through the graph neural network model, and normalize the third node feature to obtain the node stability probability corresponding to the third node feature; The second processing unit is used to obtain the stability probability corresponding to the power system by calculating the node stability probability corresponding to all nodes in the power system in a preset time period.

10. The power system stability assessment system according to any one of claims 6 or 9, characterized in that: The obtaining of the stability probability of the power system is specifically: In the formula, f( x ) is the corresponding stability probability of the power system, N is the number of nodes in the power system, N t For custom time periods, For the i The node stability probability of a node at time t.

Images